How I more than doubled my print speed while increasing accuracy

Topics around mechanical design, controller and electronics. Mods & hacks welcome.
User avatar
JohnSays
Posts: 667
Joined: Tue Apr 26, 2016 2:42 pm
Location: Saint Petersburg, FL

How I more than doubled my print speed while increasing accuracy

Postby JohnSays » Sun Mar 10, 2019 9:08 pm

Many months ago now I had a phone call with JetGuy -- the only one we ever had. It was a great pleasure and fun to talk to him and I learned a lot. One of the many things we talked about was the R3D gantry vs. a CoreXY gantry. JetGuy's opinion (the short of it) is that CoreXY is capable of a better print than the XY R3D gantry. If I am remembering correctly, some of the reasons stated were that there is a certain amount of slop in the printhead bearings and a lot of weight in the entire moving XY print mechanism. There were other good reasons why CoreXY would be better. One is that the weight of the current gantry causes sag from its own weight. Thus it is literally impossible to print a flat object: The closer the printhead moves to the center, the more sag due to weight.

In the discussion below I am referring at all times to a single extruder and NOT dual extrusion. Dual extrusion statistics are available and I can include them in this thread if there is interest.

What I Did

I got to looking at what I could do and I ended up working about 9 months on possible solutions. I decided to scrap the CoreXY idea as being too big a change for me to implement -- both because of time and lack of knowledge. In the end I have a print mechanism that is 43% lighter than the original and 30% lighter than the original modified to use the BondTech BMG. Expressed in grams, my entire moving weight (not counting the rotating X-Y mass, is 642 grams. Stock is 1119 and with the BondTech BMG mod, is 919 grams. It is also stiffer and so has 21% less sag at the center.

How did I do this? I replaced the 8mm steel XY rods with 10mm anodized aluminum rods and I manufactured my own print head and into it I fit ceramic bearings. The play is roughly less than 0.001 in, but not even noticeable when moved with my hand. In essence the play is 0, but smoothness of movement is about as perfect as it can be and does not change when heated. I have experimented with several bearing configurations and decided on some very high quality (ie very tight tolerance from the manufacturer) ceramic bearings. As a note, I tried a number of Igus bearings but found that (1) the ID is not ever concentric with the OD, and the tolerance of manufacture varies too greatly. Both these issues require a good deal of time be spent hand fitting each set as two are needed for each coordinate (X and Y).

In summary, my new print gantry is:

- 43% lighter than stock
- practically speaking, no play in the printhead bearings
- 27% less sag at center

That's what experimenting is about

Note that during the experiments, I was able to get the entire moving mass down to a skinny 382 grams of moving mass. That is 66% less than stock and 58% less than the BondTech mod alone. I did this by replacing he BondTech extruder with a Zesty Sidewinder and replacing the 8mm steel rods with hollow carbon fiber rods. Note well: I had the CF rods custom made to my specs and hollow ground to the same tolerances as the aluminum. (You cannot find these CF rods anywhere in the market.) Upon receiving the rods, I check them against spec. Bearings where Igus plastic and had to be hand matched to the precision custom made holders and to each rod. So why not go with this set up vs the BondTech and aluminum rods above? Several reasons:

1. I could not find a bearing that would stand up to the CF rods. After repeated prints at 150mm/sec and acceleration of 2500, I needed to replace the bearings at about 60 hours.

2. The Zesty Sidewinder at only 108 grams (including mounting adapter and covered breakout board box) is not quit ready for primetime. I could not get consistently good prints -- they almost always had artifact patterns in the flat surfaces. Also, I was having to run the extruder pancake stepper at about .25 vref and 600 steps/mm at 1/4 microstepping. I found tuning it to be very difficult as now I had to deal with current decay. I also never felt confident that the 8-bit board was actually keeping up with the step demand.

However, the very best reason was that at 642 grams I had already hit a weight that allowed me to move the printhead faster than other elements in the N2 will allow. End result is that as the N2 is currently built and Marlin coded, it will not go any faster. For example, I can set the acceleration of the X and Y to 5000 mm/sec/sec, but it is not going to happen. Not without replacing the stepper motors and that would require a board change.

How fast Will it Print?

First, a word about limits. What limits speed is not just the limits of the N2. It is also the limits of how fast the filament will setup/cool. I found that I can print at 200mm/sec with accel of 2500 and travel of 200mm/sec. That is all fine as long as there are no fine details to slow down for -- such as a small hole. The problem seems to be that I need a slicer that is aware of when it needs to slow down to print finer details or the filament simply stays too hot and becomes messy on subsequent passes. Also, small objects like a Benchy cannot really be printed any faster for the same reason.

So what good is it? I can print reliably at 150mm/sec to get a prototype in my hands far faster than ever before. The quality is better than okay and I've got it done for myself or a client more than twice as fast. Remember it is not just the speed of the print head, it is also the more than doubled acceleration and travel. I left the jerk at 10, but have found no difference in overall print time lapse setting it at 15. IMO, 15 just puts extra stress on everything but gives no tangible results.

My usual print speed was 50 or 60 and is now 100 to 120. I can, in fact, print the same objects with better quality twice as fast.

Other Advantages

About the new printhead. It is, of course, similar to Raise3D's, but offers these advantages:

1) It is shorter top to bottom and fits a native E3D hot end or two
2) It is lighter by 230 grams
3) It has M3 mounting threads front and back. 8 in total.
4) It holds superior precision bearings

I also redesigned breakout board case. It is now mounted horizontally and, most importantly, secures the flat ribbon cable into place. Having the flat ribbon pop out during a long print is one of the frustration some of us have experienced.

Pictures

IMG_9724.JPG


IMG_9739.JPG


IMG_9708.JPG


I'll post more about this later if anyone has an interest.
Last edited by JohnSays on Sun Mar 10, 2019 11:10 pm, edited 2 times in total.
- John

2 Raise3D N2 Duals, Bondtech BMGs, adjustable table, Panucatt SD2224 drivers, run-out sensor, thermal overload protection, Firmware 1.1.9ABH - with Lin_Advance, Palette 2 Pro, Custom E3D hot end and ultra-light carriage and printer head

User avatar
JohnSays
Posts: 667
Joined: Tue Apr 26, 2016 2:42 pm
Location: Saint Petersburg, FL

Re: How I more than doubled my print speed while increasing accuracy

Postby JohnSays » Sun Mar 10, 2019 10:38 pm

Short YouTube video of it printing here: https://youtu.be/ZT95Ky0dSuk
- John

2 Raise3D N2 Duals, Bondtech BMGs, adjustable table, Panucatt SD2224 drivers, run-out sensor, thermal overload protection, Firmware 1.1.9ABH - with Lin_Advance, Palette 2 Pro, Custom E3D hot end and ultra-light carriage and printer head

Ari
Posts: 31
Joined: Fri Dec 23, 2016 3:18 pm

Re: How I more than doubled my print speed while increasing accuracy

Postby Ari » Sun Mar 10, 2019 11:56 pm

Wow John, that is pioneering at its best.

Was that video in real time?

What size nozzle were you using?

Did you mill the extruder carrier out of aluminium or is it some other material?

I and I'm sure others on the forum, are very keen to know more. It would be nice to see photos of the surface quality that you able to achieve.

Would you ever consider selling an upgrade kit?

+ Looks professional too, very well done, am impressed.

Ari

User avatar
JohnSays
Posts: 667
Joined: Tue Apr 26, 2016 2:42 pm
Location: Saint Petersburg, FL

Re: How I more than doubled my print speed while increasing accuracy

Postby JohnSays » Mon Mar 11, 2019 4:13 am

Thanks Ari for the kind words.

The nozzle was 0.6 and the filament was R3D PLA. Yes that was in real time. When I first started printing at these speeds it was hard to watch the printhead move. It was too darn fast and just felt wrong. Now I am used to it. Actually the printhead body is made from polycarbonate with a carbon fiber mix -- not the usual CF dust, but actual fiber. It is insanely light and strong and stands up to high heat. I will be getting more pictures up to show quality at different speeds. I'm headed out of town with my daughter to go look at colleges she is interested in during the Spring Break, so I won't have time to post for another week minimum.

In terms of real pioneering work, I think that title goes to people like JetGuy and Tech2C.

As far as a kit goes, it would be pretty extensive. I haven't actually brought up the support work I did to make this happen -- small change in Marlin using ABH's 1.1.9, upgraded stepper drivers, cooling the motion board with outside air, adjusting the stepper voltage up a bit -- .6 on the vref right now, and so on. Anyway, I plan to get to all of that as long as others have an interest in the project.

I am considering going back to the Igus bearing despite their finickiness as they truly do not need any lubrication. My ceramic bearings need one drop of PTFE "oil" every few hours and I find that annoying. But I love how darn precise and just plain perfect they are otherwise.
- John

2 Raise3D N2 Duals, Bondtech BMGs, adjustable table, Panucatt SD2224 drivers, run-out sensor, thermal overload protection, Firmware 1.1.9ABH - with Lin_Advance, Palette 2 Pro, Custom E3D hot end and ultra-light carriage and printer head

User avatar
ocelot27
Posts: 41
Joined: Sat Feb 02, 2019 4:39 pm
Location: Malvern, PA

Re: How I more than doubled my print speed while increasing accuracy

Postby ocelot27 » Mon Mar 11, 2019 10:46 pm

You could go with a Mosquito hot end and loose even more weight:

https://www.sliceengineering.com/shop/t ... ito-hotend

Nothing matches it in weight, size or efficiency of the heat break - and you can change nozzles with one hand.

-john

User avatar
JohnSays
Posts: 667
Joined: Tue Apr 26, 2016 2:42 pm
Location: Saint Petersburg, FL

Re: How I more than doubled my print speed while increasing accuracy

Postby JohnSays » Mon Mar 11, 2019 11:39 pm

ocelot27 wrote:You could go with a Mosquito hot end and loose even more weight


Looks very interesting. I may indeed design a printhead to use that hot end.

Thank you john
- John

2 Raise3D N2 Duals, Bondtech BMGs, adjustable table, Panucatt SD2224 drivers, run-out sensor, thermal overload protection, Firmware 1.1.9ABH - with Lin_Advance, Palette 2 Pro, Custom E3D hot end and ultra-light carriage and printer head

User avatar
ocelot27
Posts: 41
Joined: Sat Feb 02, 2019 4:39 pm
Location: Malvern, PA

Re: How I more than doubled my print speed while increasing accuracy

Postby ocelot27 » Tue Mar 12, 2019 4:06 pm

Not sure how the N2 hot ends mount but I've already designed a 6061-T6 adapter for the Pro2 - it will take about 10mm off the z build height but I'm usually making small parts so I don't care. Looks like I will have to mod the z end stop adjustment to reach higher too. Can't wait to get my favorite nozzles back.

Also replaced the anemic part cooling "fans" with blowers that work much better - I could bridge all the way across the plate if I wanted. I have to get that project up on thingiverse...

You don't really get a sense for how tiny the thing is until you see it...
IMG_1428.jpg


-john

zemlin
Posts: 412
Joined: Sat Oct 21, 2017 2:02 pm

Re: How I more than doubled my print speed while increasing accuracy

Postby zemlin » Tue Mar 12, 2019 7:00 pm

Excellent work on the gantry mod.

User avatar
JohnSays
Posts: 667
Joined: Tue Apr 26, 2016 2:42 pm
Location: Saint Petersburg, FL

Re: How I more than doubled my print speed while increasing accuracy

Postby JohnSays » Wed Mar 13, 2019 2:33 pm

Thanks zemlin
- John

2 Raise3D N2 Duals, Bondtech BMGs, adjustable table, Panucatt SD2224 drivers, run-out sensor, thermal overload protection, Firmware 1.1.9ABH - with Lin_Advance, Palette 2 Pro, Custom E3D hot end and ultra-light carriage and printer head

zemlin
Posts: 412
Joined: Sat Oct 21, 2017 2:02 pm

Re: How I more than doubled my print speed while increasing accuracy

Postby zemlin » Fri Mar 15, 2019 11:31 pm

Since you're using IGUS bushings, I'm guessing you also got your aluminum shaft material from IGUS. I'm considering following your lead and doing a similar rod swap and gantry redesign. Was wondering if you considered the hollow 12mm material? Higher moment of inertia and lower weight than 10mm solid. Thinking about trying to get it done with the hollow 12mm rods and the JUM-02 bushings - would probably have to trim some of the bushings so they don't overlap at the rod cross-over.

User avatar
JohnSays
Posts: 667
Joined: Tue Apr 26, 2016 2:42 pm
Location: Saint Petersburg, FL

Re: How I more than doubled my print speed while increasing accuracy

Postby JohnSays » Sat Mar 16, 2019 1:18 am

Zemlin, you're not going to be able to go to 12mm and hollow. It will be too soft and too large. The igus bearings need a holder to be pressed into, and that holder takes up some space as well. Additionally you have to get the hot end situated and filament to it in a straight line. I found the absolute maximum I could go was just under 16mm for the bearing with holders. Re-read what I said about Igus bearings. I machine my own holders and then they are reamed to a perfect fit for pressing a bearing into. I spent days working that bit of machining out to get a Goldilocks fit. Next, if you go hollow, how do you attach the rods to the travelers? I turn the ends down to 8mm. Lastly, you don't need less moment. Again, read what I wrote. I was using hollow carbon fiber. It had no sag, weighed 1/2 the aluminum rods weight and got me no performance improvement.

I did extensive testing of a variety of rods using a digital force gauge. I found aluminum rods that gave less sag and some that gave more sag under the new weight than the 8mm steel did under the old weight. In the end, 10mm aluminum is not going to be stiffer than the R3D steel at the same load. CF is, but, as you read, has problems with bearings. There are aluminum rods out there with beautiful surfaces that are also too soft. I tested some of those too. Igus's work, but you have to reduce the overall weight or you could end up with more sag than with the 8mm rods.

P1020444.jpg
Testing a Carbon Fiber Rod
Last edited by JohnSays on Sat Mar 16, 2019 2:43 am, edited 1 time in total.
- John

2 Raise3D N2 Duals, Bondtech BMGs, adjustable table, Panucatt SD2224 drivers, run-out sensor, thermal overload protection, Firmware 1.1.9ABH - with Lin_Advance, Palette 2 Pro, Custom E3D hot end and ultra-light carriage and printer head

User avatar
JohnSays
Posts: 667
Joined: Tue Apr 26, 2016 2:42 pm
Location: Saint Petersburg, FL

Re: How I more than doubled my print speed while increasing accuracy

Postby JohnSays » Sat Mar 16, 2019 2:42 am

Printing Hogwarts @ 150mm/sec: https://youtu.be/LrFsZgbx49k

A few more pictures:

P1020501.RW2.jpg
Reworked breakout box and cover


IMG_9442.JPG
View of flat wire holder (the tab) in cover


P1020499.RW2.jpg
Zesty Sidewinder mounted on adapter on prototype print block
- John

2 Raise3D N2 Duals, Bondtech BMGs, adjustable table, Panucatt SD2224 drivers, run-out sensor, thermal overload protection, Firmware 1.1.9ABH - with Lin_Advance, Palette 2 Pro, Custom E3D hot end and ultra-light carriage and printer head

User avatar
ocelot27
Posts: 41
Joined: Sat Feb 02, 2019 4:39 pm
Location: Malvern, PA

Re: How I more than doubled my print speed while increasing accuracy

Postby ocelot27 » Sat Mar 16, 2019 6:11 pm

My experience with Igus bearings is that they have to be fitted perfectly with just the right amount of compression to make them perfect - you can't just throw them in any old mount - especially a 3d printed one - it will work but there will be slop. Theres also a very fine line between compressing the bearing in its mount to take up slop and them not working as "bearings" but seizing. I've had some success making slightly tight mount and then working the bearing over a rod to loosen it up - a lot of work for little reward IMO. They are also very intolerant of any weird loading on the bearing - they will seize if movement forces aren't parallel to the direction of travel.

-john

User avatar
JohnSays
Posts: 667
Joined: Tue Apr 26, 2016 2:42 pm
Location: Saint Petersburg, FL

Re: How I more than doubled my print speed while increasing accuracy

Postby JohnSays » Sat Mar 16, 2019 11:36 pm

Yes. That is correct and why I spent so much time working out a system to produce a holder that would work each time. One of the main tricks with the Igus bearings is to measure the thickness around the circumference. It will vary, but one side will always be thicker than the side 180 degrees opposite. So 0.98mm vs. 1.02mm. Mark the thick side and press it into the holder lined up with the thick side of the other bearing it will be paired with. I put a line down the outside of my holder to align the mark on the bearing to. If this is not done, I cannot even insert my rod through the two bearings. The fit needs to be that tight or it's no good. It needs to move smoothly, but catch slightly on any high spots on the rod. The bearing will settle in with time and will stop catching on the high spots.

Also, as you say john, the bearing does seize if the forces are not perfectly perpendicular to the direction of movement. In fact, when using them on the CF rods, as the bearings wore and a very slight tip of the printhead developed, it was enough to seize the movement and create skipped steps = end of print.

That is why I have been experimenting with ceramic (Rockwell 70) bearings.
- John

2 Raise3D N2 Duals, Bondtech BMGs, adjustable table, Panucatt SD2224 drivers, run-out sensor, thermal overload protection, Firmware 1.1.9ABH - with Lin_Advance, Palette 2 Pro, Custom E3D hot end and ultra-light carriage and printer head

User avatar
JohnSays
Posts: 667
Joined: Tue Apr 26, 2016 2:42 pm
Location: Saint Petersburg, FL

Re: How I more than doubled my print speed while increasing accuracy

Postby JohnSays » Sun Mar 17, 2019 12:04 am

I am now thinking that I will not try the Mosquito. My E3D is 42mm tall and the Mosquito is 41mm. The weight of the E3D with heater, fan, TC and wires is 94g. It is more weight than the original R3D at 73g, but the trade-up is worth the weight.

To try the Mosquito means creating another printhead, buying new nozzles and then dialing it in. I've spent way too much time on this present mod and need to get back to printing. If some day I need a hotter hot end, then the Mosquito may be the way to go.

I genuinely appreciate you bringing it to my attention john.
- John

2 Raise3D N2 Duals, Bondtech BMGs, adjustable table, Panucatt SD2224 drivers, run-out sensor, thermal overload protection, Firmware 1.1.9ABH - with Lin_Advance, Palette 2 Pro, Custom E3D hot end and ultra-light carriage and printer head


Return to “Hardware”

Who is online

Users browsing this forum: No registered users and 2 guests